Breast cancer is a difficult disease to manage because it is comprised of a wide spectrum of tumor subtypes with different biological characteristics, therapeutic responses and clinical outcomes. A biomarker-based classification of breast cancer that effectively matches intrinsic biological characteristics with the most effective therapeutic protocol would be a significant advance. Gene expression analysis using a breast tumor "intrinsic" gene set has reproducibly identified five distinct subtypes of breast tumors: Luminal A (LumA), Luminal B (LumB), Normal Breast-like (NB), HER2-positive (HER2+) and Basal-like. Each subtype has a distinct biology and clinical behavior and evidence suggests that each has a unique drug sensitivity pattern. We have also shown that this classification identifies prognostic groups that are reproducible across different patient populations and are independent of standard clinical parameters. RNA expression profiling is the most robust and reproducible way to identify these biological subtypes and we believe that our proposed classification can be accomplished in an automated fashion without subjective interpretation. In order to generalize the clinical significance of these findings to larger populations we will develop an assay to allow classification from formalin-fixed, paraffin-embedded (FFPE) tissues so that RNA from aged blocks can be accurately profiled. Next, we will retrospectively validate our "intrinsic" gene set on uniformly treated cohorts of thousands of patients. Our goal is to develop a clinical test for these five subtypes using expression analysis by real-time quantitative RT-PCR. With the additional information that we expect to gain by profiling clinical samples from homogeneously treated patients and patients subjected to treatment randomization, we will be able to provide valuable prognostic information for patients with node negative breast cancer and predictive information for the efficacy of chemotherapy regimens for patients with node positive disease. It is our long term goal to develop a broadly applicable subtyping test for all early stage breast cancer patients.